Review timing for processing large networks

[bdemchak]

Processing a small network is quick by interactive standards. Eric did preliminary profiling for a larger network and found bottlenecks that can be addressed. Be sure to present the timings to Dan for discussion and to determine where optimization/streamlining should be invested in.

[ericsage]

I've addressed several bottlenecks this week in the service and the client. The client now spends less time doing nothing, it now prepares resources while awaiting an async response from the server (but still protects against timeouts and other service errors). On the service I've added additional RAM and have started using a C extension JSON parser, along with reducing the amount of time spent converting between formats via throwing away unnecessary information. I think that the next step would be to implement gzip from the server to the client. After that, looking more into how we can transport less CX to the service from Cytoscape.

[bdemchak]

@ericsage made a number of suggestions that keeps this from being closed. Let's get timings under controlled circumstances so we can know where to put efforts and when we have a satisfactory result. Especially, let's look into sending less CX.